We are currently working toward automatic retrievals of crop production using satelite-based remote sensing products. Our approach involves the adaptation of the MOD17 algorithm (GPP and NPP) for agricultural applications. Higher-level MODIS products are provided as 8-day composites, but at low spatial resolution. We use statistical methods to blend MODIS and Landsat7/8 images to obtain 8day NDVI composites at 30 m resolution to determine crop phenology and fpar.
Earlier results indicate that our methods are suficiently accurate to capture county-scale production averages adn variability, but additional work is necessary to obtain accurate crop retrievals at higher resolutions.
Our current remote sensing estimates of crop production at county-scale in Montana are given below. This map is showing ongoing work and will be updated as our algorithms improve.